Objective: To determine the effect of albumin administration on morbidity in acutely ill hospitalized patients.
Data source: Computer searches of MEDLINE, EMBASE, and the Cochrane Library; hand searches of journals and Index Medicus; inquiries with investigators and fluid product suppliers; and examination of reference lists. No language or time period restrictions were adopted.
Study selection: Randomized, controlled trials comparing the administration of albumin with that of crystalloid, no albumin, or lower-dose albumin.
Data extraction: Two investigators independently extracted data. The primary endpoint for the meta-analysis was morbidity, defined as the incidence of complications, including death. Trial quality was evaluated by blinding, allocation concealment, presence of morbidity as a study endpoint, and individual patient crossover.
Data synthesis: Seventy-one trials were included in the categories of surgery or trauma, burns, hypoalbuminemia, high-risk neonates, ascites, and other indications. The 3,782 randomized patients in the included trials experienced a total of 3,287 complications, including 515 deaths and 2,772 cardiovascular, gastrointestinal, hepatic, infectious, renal, respiratory, and other complications. Albumin significantly reduced overall morbidity, with a risk ratio of 0.92 (confidence interval [CI], 0.86-0.98). Control group albumin dose significantly affected the incidence of complications (p = .002). In 32 trials with no albumin administered to the control group, the risk ratio was 0.77 (CI, 0.67-0.88) compared with 0.89 (CI, 0.80-1.00) in 20 trials with control patients receiving low-dose albumin and 1.07 (CI, 0.96-1.20) in 19 trials with moderate-dose control group albumin.
Conclusions: Albumin reduces morbidity in acutely ill hospitalized patients. Concomitant administration of albumin in the control group can obscure the effects of albumin on clinical outcome in randomized trials.